Structural Determination of a Short-lived excited Iron(II) Complex by Picosecond X-ray Absorption Spectroscopy

(1)

Article

Reference

Structural Determination of a Short-lived excited Iron(II) Complex by Picosecond X-ray Absorption Spectroscopy

GAWELDA, Wojciech, et al.

Abstract

Structural changes of the iron(II)-tris-bipyridine ([FeII(bpy)3]2+) complex induced by ultrashort pulse excitation and population of its short-lived (0.6 ns) quintet high spin state have been detected by picosecond x-ray absorption spectroscopy. The structural relaxation from the high spin to the low spin state was followed over the entire lifetime of the excited state. A combined analysis of the x-ray-absorption near-edge structure and extended x-ray-absorption fine structure spectroscopy features delivers an Fe-N bond elongation of 0.2 Å in the quintet state compared to the singlet ground state.

GAWELDA, Wojciech, et al . Structural Determination of a Short-lived excited Iron(II) Complex by Picosecond X-ray Absorption Spectroscopy. Physical Review Letters , 2007, vol. 98, no. 8, p. 57401

DOI : 10.1103/PhysRevLett.98.057401

Available at:

http://archive-ouverte.unige.ch/unige:3198

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

Structural Determination of a Short-Lived Excited Iron(II) Complex by Picosecond X-Ray Absorption Spectroscopy

Wojciech Gawelda,^1,2Van-Thai Pham,¹Maurizio Benfatto,³Yuri Zaushitsyn,¹Maik Kaiser,²Daniel Grolimund,² Steven L. Johnson,²Rafael Abela,²Andreas Hauser,⁴Christian Bressler,^1,*^,†and Majed Chergui^1,*^,‡

1Ecole Polytechnique Fe´de´rale de Lausanne, Laboratoire de Spectroscopie Ultrarapide, ISIC, FSB-BSP, CH-1015 Lausanne, Switzerland

2Paul Scherrer Institut, Swiss Light Source, CH-5232 Villigen PSI, Switzerland

3Laboratori Nazionali di Frascati, INFN, CP13, I-00044 Frascati, Italy

4De´partement de Chimie Physique, Universite´ de Gene`ve, 30 quai Ernest-Ansermet, CH-1121 Gene`ve, Switzerland (Received 2 August 2006; published 2 February 2007)

Structural changes of the iron(II)-tris-bipyridine (Fe^IIbpy₃²) complex induced by ultrashort pulse excitation and population of its short-lived (0:6 ns) quintet high spin state have been detected by picosecond x-ray absorption spectroscopy. The structural relaxation from the high spin to the low spin state was followed over the entire lifetime of the excited state. A combined analysis of the x-ray- absorption near-edge structure and extended x-ray-absorption fine structure spectroscopy features delivers an Fe-N bond elongation of 0.2 A˚ in the quintet state compared to the singlet ground state.

DOI:10.1103/PhysRevLett.98.057401 PACS numbers: 78.70.Dm, 82.40.g, 82.53.Kp

Molecular iron(II) complexes exhibit strongly coupled electronic, magnetic, and structural dynamics. They have been extensively studied in relation to the phenomenon of spin crossover (SCO), where a transition from a low spin (LS) ground state to a high spin (HS) excited state is induced by either temperature or light [1], while the reverse transformation can be induced by pressure [2]. The study of SCO compounds is being actively pursued for potential applications in magnetic data storage and as a basis for bistable devices [3]. In biology, SCO plays an important role in the binding of ligands in heme proteins [4]. Finally, ultrafast spin changes (like the case presented here) may offer a route for data processing at the nanoscale [5].

A characteristic energy level scheme for iron(II) complexes is shown in Fig.1. In compounds with a sufficiently low zero-point energy difference E⁰_HL between LS and HS states, i.e., of the order of100–1000 cm¹, SCO can be temperature induced, but this is not possible for FeÎIbpy₃² with a E⁰_HL6000 cm¹. The population of the HS state of Fe(II)-based SCO complexes can be triggered optically in the so-called light-induced excited spin state trapping (LIESST) process [1,6]. However, the lifetime of the HS state of FeÎIbpy₃² is limited to microseconds at cryogenic temperatures, in contrast to the light-induced HS state of SCO complexes with low- temperature lifetimes of more than 10 h. Photoexcitation of FeÎIbpy₃² by UV-visible light populates the singlet metal-to-ligand-charge-transfer (¹MLCT) states, and is followed by a cascade of intersystem crossing steps through MLCT and ligand-field (LF) states, which brings the system to its lowest-lying quintet state (HS),⁵T₂, with almost unit quantum yield within <1 ps [7,8]. At room temperature, this state relaxes nonradiatively to the LS ground state in 0:6 nsin aqueous solutions. Knowledge of the relationship between the HS!LS relaxation rate,

E^y_HL, and the Fe-N equilibrium distance is crucial for the design of molecules with longer-lived HS states.

Steady-state structural studies on Fe(II) SCO compounds, both in the LS ground and the HS states gener- ated by LIESST or temperature, have been carried out using x-ray crystallography and x-ray absorption spectroscopy (XAS), all delivering an elongation of the Fe-N bond by 0.18– 0.22 A˚ , with respect to the LS ground state

FIG. 1. Schematic potential energy curves of Fe(II)-SCO complexes as a function of the Fe-N bond distance, and pathways of nonradiative relaxation leading to the short-lived quintet state (0.6 ns forFe^IIbpy₃²), upon excitation of the singlet MLCT state in the UV.

(3)

[9–21]. Recently, Fe K-edge steady-state XAS of the LS Fetrenpy₃PF₆₂ and its HS analogue Fetren6-Me-py₃PF₆₂ were combined to interpret the picosecond transient spectra of the photoactivated Fetrenpy₃PF₆₂ complex, showing good agreement with the anticipated result for the HS complex [22].

This Letter presents a picosecond study at the FeK-edge absorption of photoexcited aqueous Fe^IIbpy₃². The key point here is that, contrary to LIESST-SCO complexes, for Fe^IIbpy₃² picosecond XAS is the only possible method to determine the structural parameters of the light- induced HS state, because its lifetime is too short for quasistatic structural studies, even at the lowest temperatures. Here, we capture the structure of the short-lived HS intermediate (0:6 ns), with 100 ps hard x-ray pulses, and map out the evolution of the structure over the entire HS lifetime. In addition, and contrary to previous work [22], we extract the structural parameters of the HS state directly from a fit of the experimental difference absorption spectra without prior assumptions about its structure. The structural parameters of the HS state were retrieved from both the x-ray absorption near edge structure (XANES) [23,24] and the extended x-ray absorption fine structure (EXAFS) [25], delivering identical results within their uncertainties.

The laser-pump x-ray-probe experiments were per- formed at the micro-XAS beam line of the Swiss Light Source. Details of the experimental strategy are described elsewhere [26–28]. Briefly, an intense 400 nm pulse (100 fs pulse width, repetition rate 1 kHz) excites an aqueous solution of 1– 25 mM Fe^IIbpy₃², while a 100 ps monochromatic (0.016% bandwidth) tunable hard x-ray pulse probes the system by x-ray absorption spectroscopy as a function of the adjustable pump-probe time delay. The detected signal is the difference x-ray absorption spectrum between the laser excited and the unexcited sample, recorded on a shot-to-shot basis at 2 kHz. The XAS of the sample was measured in both transmission and x-ray fluorescence yield modes. The incident x-ray flux was monitored via x-ray fluorescence from a thin Cr foil placed upstream from the sample. The x-ray energy was calibrated via a reference spectrum of an iron foil.

For the quantitative structural analysis, we used two approaches. First, we implemented the MXAN code described in Refs. [23,24]. It uses the so-called full multiple scattering (FMS) approach, avoiding anya prioriselection of the relevant multiple scattering paths, together with the muffin-tin approximation for the shape of the atomic po- tentials [29], which are recalculated and optimized at each step of the least-squares minimization procedure. Here, we use a recent extension of the MXAN code [30] to fit the difference x-ray absorption spectra in energy space, which is more sensitive for the structural analysis. In the second approach, we fitted the Fourier-transformed EXAFS spectra of the LS and HS spectra using the theoretical scattering amplitudes and phases delivered by theFEFF 8.2code, after

applying the appropriate data reduction procedure to con- vert the spectra to wave vector space [31].

Figure 2(a)shows the static XAS spectrum of the first 200 eV of the FeKabsorption edge ofFe^IIbpy₃². It is characterized by a number of XANES features (labeled A to D), whose assignments were already discussed in Refs. [16,32]. The resonances above 50 eV from the edge are predominantly single scattering features (i.e., the EXAFS), in particular, feature E, which is dominated by scattering from the nearest N atoms. In XAS studies of similar compounds it was observed that all these features change drastically upon SCO [16,19–21], as seen in Fig.2(b)for the transient absorption spectrum (50 ps after excitation) and in the reconstructed (as explained below) HS spectrum [Fig.2(a)]. Most of the changes at theBtoE bands are attributed to changes of metal-ligand and intra- ligand bond distances and angles, with the Fe-N bond being the dominant contributor. Additional changes in the high energy (EXAFS) region (not shown here) are also clearly observed, which likewise point to a significant

0.0 0.5 1.0

0 50 100 150 200

0.0 0.1

0 20

D

Absorption / a.u.

a)

b)

E

C D B

Transient Absorption / a.u.

X-Ray Probe Energy /eV

A

FIG. 2. (a) XANES spectrum of the LS state of a 25 mM aqueous solution ofFe^IIbpy₃²(circles) and simulatedMXAN

spectrum (solid line) using an Fe-N bond distance of 2.0 A˚ . The inset shows details of the edge region and of the fit. The dots represent the XANES spectrum of the HS state retrieved via Eq. (1). The spectrum is plotted with respect toE0(7122.5 eV).

(b) Experimental difference XANES recorded 50 ps after laser excitation (circles), and the fit results obtained by the FMS theory (solid line) with an Fe-N bond distance of 2.0 A˚ and a chemical shift of2:5 eV(see text for details).

057401-2

(4)

Fe-N bond change. Figure 3compares the temporal evolution of the absorption changes at the B feature (7126 eV) with the kinetics of the ground state recovery measured by femtosecond optical pump-probe spectroscopy [8]. The latter reflects the repopulation of the LS state by the decay of the HS state and it matches the x-ray data perfectly.

Our results differ from previous XAS studies of light- induced SCO systems, which used exclusively the LIESST phenomenon in low-temperature solids to gain information about the structural characteristics of the HS state [16,18,20,21]. Also in Ref. [22], the detected HS complex in solution was relatively long-lived (60 ns), compared to the experimental resolution of the experiment (70 ps).

Here, we impulsively photoinduce a transition to the HS state in a room temperature solution using a femtosecond laser pulse and follow the evolution of its electronic and structural properties over its entire sub-ns lifetime.

The difference absorption spectrum [Fig.2(b)] is defined as [33]

AE;t ftA_HSE;t A_LSE; (1) whereftis the fractional population of the HS complex at time delay t [50 ps in Fig. 2(b)], A_LSE is the absorption spectrum of the LS complex [Fig. 2(a)], and A_HSE;tis that of the HS complex, at the time delayt after photoexcitation. In order to extract the excited state structure correctly,ftmust be known, and we measured a value of 222% at t50 ps, in laser-only pump- probe experiments carried out under identical experimental conditions (laser fluence, sample concentration, and thick-

ness) as in the x-ray experiments. The fractional HS population was estimated from the bleach of the ground state absorption upon 400 nm excitation [8], taking into account the unit quantum yield for populating the quintet state [7].

The recovered HS state XAS spectrum is shown in Fig.2(a).

For the structural analysis, we start off by fitting the XANES spectrum of the LS ground state using theMXAN

code [Fig.2(a)], as a check of the initial parameters. The fit [Fig. 2(a)] agrees very well with the experimental spectrum, including the edge region (see inset), delivering an Fe-N bond length of R_Fe-N2:000:02 A, in good agreement with crystallographic data (1:9670:006 A) [34] and density functional theory (DFT) calculations (1:990:02 A) [35]. Next, we fit the difference spectrum [Fig. 2(b)] to extract the Fe-N bond distances in the HS state. This is done by simulating the XANES spectrum of the HS state for varying Fe-N bond distances, and taking the difference with the LS state. However, the bond changes also modify the electronic structure, affecting both the edge features (A and B), the ionization potential E₀7122:5 eV, and features C, D, and E, above it. This is rationalized by the chemical shift, which represents the shift of the ionization potential E₀ of the Fe atom, due to the bond length changes in the HS state.E₀ is critical for the structural determination. Since its shift is not known a priori, we carried out a Monte Carlo optimization of the fit by minimization of the square residual function [30], which measures the quality of the agreement between the experimental and simulated traces. In the first optimization, we took the experimental photoexcitation yield of 22%, and obtained a chemical shift E₀ 2:5 0:5 eVand an Fe-N bond elongation ofR_Fe-N0:20 0:05 A. As a confirmation, we then fixedE₀to2:5 eV and repeated the optimization with ft50 psas adjustable parameter, yielding ft50 ps 20%–23%

and R_Fe-N0:190:03 A. The resulting fit curve is presented in Fig. 2(b), showing good agreement with the experimental difference spectrum.

Finally, we also fitted thek³-weighted EXAFS spectra of the LS and HS complexes using the scattering amplitudes and phases obtained with theFEFF 8.20code. Figure4shows a comparison of the Fourier-transformed experimental and fitted spectra (not corrected for the central atom phase shift, which corresponds to 0.3 A˚ ). The fits were obtained using the first-shell contribution only and both LS and HS Fe-N bond distances were freely optimized in the fits (re- strained to all 6 N atoms having the same bond distance).

The LS bond distance is1:990:02 A, while for the HS state we find R_HS2:190:04 A, i.e.,R_Fe-N0:2 0:02 A, fully consistent with the above MXAN results.

Important in the EXAFS fit is the value of the chemical shift, since it determines the value of the photoelectron wave vector. We optimized it independently and obtained a value of E₀ 2:80:5 eV for the HS state, which agrees remarkably well with theMXANresult.

-200 0 200 400 600 800 1000 0

1 Transient Absorption / a.u.

XAFS Optical

Time Delay / ps

FIG. 3. Kinetics of the difference x-ray absorption fine structure (XAFS) signal of aqueousFe^IIbpy₃² at room temperature, recorded at 7126 eV (circles) upon 400 nm excitation, and that of the optical signal (dots) recorded in transmission at 523 nm, both reflecting the population of the excited state. The solid lines represent monoexponential fit curves with a decay constant of 650 ps. In the case of the x-ray signal, convolution with the 100 ps width of the x-ray pulse was included. Both spectra and fit curves are normalized to half-height of the rise.

(5)

The R_Fe-Nvalues of 0:2 A extracted from both the XANES and the EXAFS are in very good agreement with the LS-to-HS bond elongation from DFT calculations on Fe^IIbpy₃² [35], and with the values derived for other Fe(II)-based SCO complexes using time resolved [22] and static structural methods [9–18,20,21]. That the bond elongation is nearly the same for all complexes, despite their largely different HS lifetimes, confirms the fact that the driving force for the HS-LS relaxation is determined by the energetics of the HS state, rather than its structure [36].

The time scale for the relaxation steps from the singlet

1MLCT to the HS state is subpicosecond [8]. Optical spectroscopy is unable to resolve them, as they are spec- troscopically silent [8]. Future extensions using femtosecond x-ray pulses extracted from a synchrotron, using the so-called ‘‘slicing’’ scheme [37], will soon provide insight into these steps and the structural changes therein occurring.

This work was funded by the Swiss National Science Foundation (FNRS), via Contracts No. 620-066145, No. 200021-107956, and No. 200021-105239. We thank M. Willimann, M. Harfouche, and B. Meyer for their assistance and help during the measurements.

*Corresponding author.

†Electronic address: Christian.Bressler@epfl.ch

‡Electronic address: Majed.Chergui@epfl.ch

[1] P. Gu¨tlich, A. Hauser, and H. Spiering, Angew. Chem., Int.

Ed. Engl.33, 2024 (1994).

[2] V. Ksenofontov, A. B. Gaspar, and P. Gu¨tlich, in Spin Crossover in Transition Metal Compounds III, Topics in

Current Chemistry Vol. 235, edited by P. Gu¨tlich and H. A.

Goodwin (Springer, Berlin, 2004), p. 23.

[3] J. F. Letard, P. Guionneau, and L. Goux-Capes, in Spin Crossover in Transition Metal Compounds III(Ref. [2]), p. 221.

[4] J. N. Harvey, Faraday Discuss.127, 165 (2004).

[5] G. P. Zhanget al., inSpin Dynamics in Confined Magnetic Structures I, Topics in Applied Physics Vol. 83, edited by B. Hillebrands and K. Ounadjela (Springer, Berlin, 2002), p. 245.

[6] A. Hauser, in Spin Crossover in Transition Metal Compounds II, Topics in Current Chemistry Vol. 234, edited by P. Gu¨tlich and H. A. Goodwin (Springer, Berlin, 2004), p. 234.

[7] C. Brady et al., in Spin Crossover in Transition Metal Compounds III(Ref. [2]), p. 1

[8] W. Gaweldaet al., J. Am. Chem. Soc. (to be published).

[9] J. Kusz, H. Spiering, and P. Gu¨tlich, J. Appl. Crystallogr.

34, 229 (2001).

[10] E. J. MacLeanet al., Chem. Eur. J.9, 5314 (2003).

[11] J. Kuszet al., J. Appl. Crystallogr.38, 528 (2005).

[12] A. L. Thompsonet al., Chem. Commun. (Cambridge) 12 (2004) 1390.

[13] K. Okamotoet al., Chem. Phys. Lett.371, 707 (2003).

[14] P. Guionneauet al., inSpin Crossover in Transition Metal Compounds II(Ref. [6]), p. 97.

[15] M. Hostettleret al., Angew. Chem., Int. Ed. Engl.43, 4589 (2004).

[16] C. Hannayet al., Inorg. Chem.36, 5580 (1997).

[17] P. J. van Koningsbruggen et al., Inorg. Chem. 39, 1891 (2000).

[18] T. Yokoyamaet al., Phys. Rev. B58, 14 238 (1998).

[19] M. L. Boillotet al., New J. Chem.26, 313 (2002).

[20] H. Oyanagi, T. Tayagaki, and K. Tanaka, J. Phys. Chem.

Solids65, 1485 (2004).

[21] H. Oyanagi, T. Tayagaki, and K. Tanaka, J. Lumin.119, 361 (2006).

[22] M. Khalilet al., J. Phys. Chem. A110, 38 (2006).

[23] M. Benfattoet al., J. Synchrotron Radiat.8, 267 (2001).

[24] M. Benfattoet al., Phys. Rev. B65, 174205 (2002).

[25] A. L. Ankudinovet al., Phys. Rev. B58, 7565 (1998).

[26] M. Saeset al., Rev. Sci. Instrum.75, 24 (2004).

[27] W. Gaweldaet al., Phys. Scr.T115, 102 (2005).

[28] W. Gaweldaet al., J. Am. Chem. Soc.128, 5001 (2006).

[29] T. A. Tysonet al., Phys. Rev. B46, 5997 (1992).

[30] M. Benfattoet al., J. Phys. Chem. B110, 14 035 (2006).

[31] B. K. Teo, EXAFS: Basic Principles and Data Analysis (Springer, Berlin, 1986).

[32] V. Brioiset al., Inorg. Chem.40, 912 (2001).

[33] C. Bressler and M. Chergui, Chem. Rev.104, 1781 (2004).

[34] S. Dick, Z. Kristallogr. New Cryst. Struct. 213, 356 (1998).

[35] L. M. L. Dakuet al., Chem. Phys. Chem.6, 1393 (2005).

[36] A. Hauser, N. Amstutz, S. Delahaye, S. Schenker, A. Sadki, R. Sieber, and M. Zerara, in Optical Spectra and Chemical Bonding in Inorganic Compounds, Structure and Bonding Vol. 106, edited by Th.

Scho¨nherr (Springer, Berlin, 2003), p. 81.

[37] R. W. Schoenleinet al., Science287, 2237 (2000).

0 1 2 3 4

0 2 4 6

FT Magnitude |χ(R)| /Å-4

Fe-N Bond Length /Å

FIG. 4. Fourier Transforms (FT) ofk³-weighted EXAFS spectra of the LS (circles) and HS (squares) complexes together with their fits (solid lines) using a first-shell model with 6 nearest neighbor N atoms (FT magnitudes are not phase corrected;

therefore, they appear at shorter bond lengths).

057401-4